Prototype

[andrew]

Very early prototype development of how open collective oss funds could work:

• [ ] define the funds
  • initial list of funds to come from the featured topics from github.com/topics (indexed in ecosystems over here: https://awesome.ecosyste.ms/topics) with some filtering for very small or very broad topics.
• [ ] define the list of projects within the funds
  • initial list of projects to include within each fund will be based on the topics/keywords of each fund
  • topics that align closely with large programming language ecosystems should use ecosystem data defined at the package manager level from https://packages.ecosyste.ms/ instead of topics, or primary language (most javascript packages don't use the topic javascript for example but are detected as being written in javascript and published to npmjs.com)
  • may want to add in the most popular dependencies of projects within the fund as well
  • all projects must have a license clearly defining how it can be used, modified, redistributed and referenced by others.
  • projects will most likely defined by the source repository, rather than at an individual package level (i.e. if two matching packages in a fund are from the same git repository, they will be counted as the same project)
  • there may need to be some basic quality filters to act as a spam filter
• [ ] define the fund distribution algorithm
  • ran monthly to produce a table of projects and amount to recieve based on the value in the fund at the end of the month
  • likely to be proportional based on usage within the fund with some minimum cut-off ($50 for example)
• [ ] define the format for the fund distribution algorithm output
  • csv that open collective staff can use to distribute funds
  • use funding links where available (prefer open collective to minimize transfer costs)
  • find emails of maintainers for projects as contact details
  • group projects and their fund amounts by funding platform so only one combined payment is made to each platform each month

Other things to consider:

• [ ] database should record results of fund outputs each month and provide detailed summary of project funds over time
• [ ] dry runs of funding algorithms with different parameters would be helpful for testing purposes
• [ ] clear documentation of how a project was considered part of a fund and how other projects can propose themselves to be added to the fund in future runs.

[BenJam]

define the format for the fund distribution algorithm output

the distribution 100% needs to be automated, meaning that we need to create/update the contributions/invitations on a monthly basis.

The CSV is when we have a platform representing multiple projects. We'd like to accumulate those into a single payment, then send the platform a CSV so they have complete information about distribution their side.

[andrew]

Very early prototype is coming together, primarily focused on the fund distribution algorithm and it's inputs.

[andrew]

And a version which includes a breakdown by funding platform

[andrew]

Some more TODOs off the top of my head:

• [x] minimum amount (dollars)
• [ ] maximum amount (percentage)
• [ ] modifier for already highly funded projects
• [x] keep track of funding accounts (oc, github sponsors, etc, may also be cross-fund)
• [ ] a way of projects opting out
• [ ] preferred funding link should be possible for projects to change (except perhaps always using open collective if available)

[andrew]

breakdown by funding platform and project, recorded against each allocation run

[andrew]

opencollective platforms should be grouped by host_id

[andrew]

Idea: check what your projects payout would be for a fund

[andrew]

show total number of possible projects considered in allocation

[andrew]

How interdependent is a project within an ecosystem